The BLS uses a sample of 65,000 households which amounts to over 100,00 people for their survey this is called a CPS (current population survey). The CPS sample is selected so as to be representative of the entire population of the United States. In order to select the sample, all of the counties and county-equivalent cities in the country first are grouped into 2,025 geographic areas. For my data on population I used the census bureau's website. The census bureau uses the population estimates program (PEP) in order to gather its data. This program measures the citizens in a state that are "usually residents" this means that if people own a home or mailing address in that state (other than summer homes) they are considered a resident of that state. I decided to include a population as a regressor because I wanted to see if more people in an area meant more jobs in that area. For my data on education spending I used usgovernmentspending.com. On this website they describe the way they got their data and it seems to me to be similar to the methods of forecasting used in econometrics. They gather the data and use a 5-year reference period in order to forecast change. Since I am only using data on 2011 I didn't need to acquire the data they forecasted, they got their yearly data from the Census Bureau's report on state and local government expenditure. I decided to include education spending in this regression because I wanted to see whether or not the spending of the state on child development had anything to do with the unemployment in that state.The general consensus is that the better-educated people are the more likely they are to get a job. For my data on colleges/trade schools in each state I used univsearch.com this is a pretty simple website that lists off all of the universities and trade schools that are based out of each state. For their data they used the National Center of Education Statistics and the U.
